A model of political connections and finance
1Michael O’Connor Keefe Victoria University of Wellington School of Economics and Finance PO Box 600, Wellington 6140, NZ
[email protected] +64 04 463 5708
Current Draft: 28th September, 2015
1I thank Victoria University of Wellington for granting and funding my research leave and Willamette University for hosting me. I thank Gary Knight and Loc Phuoc Nguyen for their helpful comments. All remaining errors are my own.
A model of political connections and finance
Abstract
To model the influence of political connections on credit, I extend the fixed investment model with stochastic contract enforcement of Tirole (2006, pages 535-540). The model provides several insights including: i) political influence over the lender’s future compensation leads to loans with an expected rate of return less than 100%, ii) political influence over the borrowers pursuit of social objectives tightens the financial constraints of the politically connected, and iii) the co-existence of political connected and non-connected borrowers in the same economy can be interpreted as borrowers choosing to be politically unconnected so as to pursue high return projects and overcome credit rationing.
Keywords: Political Connections, Moral Hazard, Financial Constraints
JEL Classification Codes: G32–Financing Policy, D72 – Economic models of political processes, D86–Economics of Contract: Theory
1 Introduction
In the fixed investment model with stochastic contract enforcement of Tirole (2006, pages 535-540), the lenderex-ante adjusts the allocation of return between the lender and borrower to account for possible expropriation. Due to this ex-ante adjustment of the contracts for weak enforcement (possible expropriation), the interest rate increases, but the expected rate of return on the loan is unchanged. Empirical results suggest this adjustment for possible expropriation does not fully capture the lender’s incentives. For example, using loan level data of borrowers in Pakistan, Khwaja and Mian (2005) find the rate of return on loans to politically connected borrowers is six percent below the rate of return on loans to politically unconnected borrowers and further the average rate of return across the entire sample is 93.46%.2 In addition, they find the interest rate charged for politically connected and unconnected loans is almost identical. To explain these empirical findings, I extend Tirole’s model by adding features that account for i) the influence of a politically connected borrower on lender compensation, and ii) the possible requirement of the politically connected borrower to pursue social objectives that reduce project return.
To model the political influence on the lender, I add additional lender compensation due to political connections. The mechanics by which a politically connected borrower influences the compensation of a lender is discussed by Khwaja and Mian (2005, p. 1373) who write,
Politically powerful firms obtain rents from government banks by exercising their polit- ical influence on bank employees. The more powerful and successful a politician is, the greater is his ability to influence government banks. This influence stems from the orga- nizational design of government banks that enables politicians to threaten bank officers with transfers and removals, or reward them with appointments and promotions.
Consistent with the empirical findings of Khwaja and Mian (2005) the model shows the expected rate of return on a loan is less than 100% when lenders receive additional compensation for lending to politically connected borrowers.
2The rate of return used by Khwaja and Mian (2005) is equal to (1 - Default Rate)*(1+ Interest Rate)+ Default Rate * Recovery Rate, which essentially measures the percentage of the loan the lender collects. 74% of the total loans in the sample are with politically connected borrowers.
Tirole (2006) assumes that the enforcement probability of a contract is an institutional feature of the economy. I follow Tirole in modeling the enforcement probability, but interpret the probability of contract enforcement as borrower specific, where a loan to a politically connected borrower has a smaller probability of being enforced. This is consistent with the empirical results of Berkman, Cole, and Fu (2009), who find the probability of expropriation is related to firm characteristics. By interpreting enforcement probabilities as being negatively influenced by political connections, the model provides an explanation for finding by Khwaja and Mian (2005) that the interest rates of the politically connected and unconnected borrowers are approximately equal. Specifically, the model shows that the interest rate of the loan increases as the probability of contract enforcement decreases (expropriation risk decreases), but decreases with politically influenced lender compensation. Thus, the effects of enforcement and politically influenced lender compensation may balance each other out so that there is a zero net effect on the interest rates due to political connections.
A politically connected borrower not only has influence on the compensation of the lender, but also may be expected to contribute to social objectives. Relative to importance of social objectives in China Chen, Jiang, Ljungqvist, Lu, and Zhou (2015, page 3) write,
. . . the objective function of the Chinese Communist Party (CCP), which ultimately controls most functions of state, is not exclusively the maximization of profits or share- holder value but also the maintenance of a “harmonious society.” Consistent with this, we document that the chairmen of state groups in our sample are rewarded with pro- motions to higher political office not only for raising productivity but also for avoiding large scale job losses. Clearly, these aims can be in conflict (maintaining overstaffing may make raising productivity difficult) and over time may be incompatible (subsidizing unproductive jobs may divert resources away from creating productive ones).
In other words, a politically connected borrower may be expected to divert resources to meet social objectives. However, the diversion of resources to social objectives is not free, but is implicitly paid for through future compensation (e.g. promotions). Consistent with the diversion of resources to social objectives, Chen, Sun, Tang, and Wu (2011) show that political influence reduces investment efficiency.
Social objectives influence credit constraints. The cost of social objectives increases the required minimum project return and cash holdings required to obtain financing. First, a lender understands that a politically connected borrower allocates a portion of a projects return to social objectives, which decreases the return available to pay off the loan. Second, the cost of social objectives de- creases the motivation of the borrower to work hard, which increases the cash required to remain incentive compatible. Thus, the loss in return due to the borrower fulfilling social objectives in- creases both the return required and the cash required to obtain financing. In contrast, politically influenced lender compensation decreases the minimum required return and cash required for a borrower to obtain financing. All in all, the effect of political connections on access to financing depends on the relative magnitudes of the social objectives and influence over lender compensation.
The cost of social objectives also influences the decision of the borrower about whether to be politically connected or unconnected. The model shows that a borrower with relatively high probability of success is more likely to choose to be politically unconnected. Intuitively, in the case of a relatively high probability of success, the borrower prefers to fully capture the economic rents of the project in lieu of the gains from political connections. Finally, in the case of relatively high social cost with a high probability of project success, a politically unconnected borrower is less financially constrained than a politically connected borrower. Thus, the co-existence of political connected and non-connected borrowers in the same economy can be interpreted as borrowers choosing to be politically unconnected so as to pursue high return projects and overcome credit rationing.
An implication that borrowers with relatively higher success probabilities (i.e. lower default rates) choose to be politically unconnected is that these borrowers seek out financing from banks that are relatively independent of political influence. Consistent with the idea that politically unconnected borrowers have higher success probabilities Khwaja and Mian (2005, p. 1392) write,
Comparing average default rates for firms that (i) borrow only from government banks, (ii) borrow from both bank types, and (iii) borrow only from private banks, shows that the first have the highest average default rates (25.7 percent), followed by the second (16.9 percent), and then the last category has the lowest default rates (5.4 percent).
In addition, Wei and Zhu (2015) report that state owned firms (relative to private firms) have lower
returns on equity and suffer more frequent losses. Overall, the model predicts that borrowers with relatively high probabilities of success choose not to be politically connected is consistent with the empirical data.
My model is broadly related to theoretical papers about rent seeking through government influence. For example, Krueger (1974) analyzes the rent seeking through restraints on trade.
Shleifer and Vishny (1993) and Shleifer and Vishny (1994) explore the influence of corruption. To my knowledge, this paper is the first paper to model the effect on financing from political influence due to politically influenced lender compensation and social objectives.
The paper proceeds as follows. Section 2 sets up a simple model of investment with moral hazard, stochastic contract enforcement, politically influenced lender compensation, social objectives that decrease project return, and implicit payments to the borrower for meeting those social objectives.
Section 3 explores the implications of political connections on the contracts, interest rates, expected payoffs, and access to credit. Section 4 derives the conditions under which it is beneficial for the borrower to be politically unconnected. Section 5 concludes.
2 Model set-up
To model the influence of political connections on credit, I extend the fixed investment model with stochastic enforcement of Tirole (2006, pp. 535-540). In Tirole’s model, the entrepreneur needs to invest I att= 0 with returns at t= 1 ofR in the case of success and 0 in the case of failure. The entrepreneur has an endowment at t= 0 of cashA and seeks to borrow I−A. Cash is restricted to 0 < A < I. The risk free rate is assumed to be zero. Both the entrepreneur and lender share the same beliefs relative to all exogenous parameters in the model. The borrower and lender are risk neutral.
At t = 0 the entrepreneur chooses whether to work hard or shirk. If she works hard, the probability of project success is pH. If she shirks, the probability of success drops from pH topL so that ∆p = pH −pL > 0. If she shirks she enjoys a private benefit of B where B > 0. The entrepreneur can not pledge the private benefit B to the lender. In addition, institutions enforce contracts with probabilitye. Thus, there are three future states: success with enforcement, success
with expropriation, and failure. For example, if the entrepreneur works hard the probability of success with enforcement is epH. Also, there are competitive market for loans so that the lender enters into a contract with zero expected utility, implying a risk neutral lender earns zero NPV in expectation. Lastly, the model’s use of a single entrepreneur (the borrower) and a single lender (the bank) eliminates several layers of agency issues. For example, to the extent a lending officer doesn’t exactly internalize the zero net present value bank lending objective, a lending officer’s decision may diverge from the model.
I model the effect of political connections on financing in three ways. First, I assume the polit- ically connected entrepreneur has costly social objectives. For example, the politically connected entrepreneur might have firm employment objectives. These social objectives reduce the overall return of the project. I define the loss from implementing these social objectives in the case of project success asLwhere 0≤L < R.3 In the case of project failure, where the gross return of the project is zero, the loss due to meeting social objectives is also zero. As a member of a political network, the entrepreneur gains future earnings by meeting social objectives. I define the present value of the increase in future earnings from meeting these social objectives asS ≥0. To evaluate the relationship between L and S, I appeal to the notion of a political central planner. To the political central planner, L represents a social benefit and S a cost. Because the central planner can either spend S directly or spend S to achieve a social benefit of L, the central planner will only choose to work through the entrepreneur if S≤L. I further assume, that the borrower can’t pledge the present valueS of future compensation to the lender. The borrower receives a payment of Rb in the case where the project succeeds and the contract is enforced.
A second means of political influence happens when a politically connected borrower influences the future compensation of the lender. I defineC ≥0 as the increase in compensation to the lender due to a borrower being politically connected. I further restrict the additional compensation due to political influence to be less than the loan or 0≤C <(I−A). Due to the competitive market for loans, the lender enters into a contract with a zero expected NPV that includes the compensation due to political connections. Overall, the effect ofC is to add a payment to the lender making the
3For the project to have a positive NPV it is necessary thatL < R.
budget constraint soft.
Based on the assumptions noted above, Table 1 shows the investments and payoffs associated with the total project, the borrower and the lender. Panel A shows the total, borrower, and lender investments at t = 0. Panels B and C show payoffs for the total project, the borrower, and the lender. Panel B assumes the entrepreneur works hard and Panel C the entrepreneur shirks.
Table 1: Investments and Payoffs
This table shows the investments and payoffs associated with total project, the borrower and the lender. Panel A shows the total, borrower, and lender investments at t= 0. Panels B and C show the payoffs and expected NPVs for the total project, the borrower, and lender. Panel B assumes the entrepreneur works hard and Panel C the entrepreneur shirks.
Panel A: Investment att= 0
Total Borrower Lender
Investment I A I−A
Panel B: Cash Flow att= 1 if entrepreneur works hard
Probability Outcome Total Borrower Lender
epH Enforcement & Success R−L+S+C Rb−L+S R−Rb+C (1−e)pH Expropriation & Success R−L+S+C R−L+S C
1−pH Failure C 0 C
Panel C: Cash Flow att= 1 if entrepreneur shirks
Probability Outcome Total Borrower Lender
epH Enforcement & Success R−L+S+C+B Rb−L+S+B R−Rb+C (1−e)pH Expropriation & Success R−L+S+C+B R−L+S+B C
1−pH Failure C+B B C
Using the payoffs from Table 1, I estimate the expected total project NPVs conditional on whether the entrepreneur works hard or shirks. If the entrepreneur works hard then the expected total project NPV is
E[N P V|Hard]T otal =epH(R−L+S) + (1−e)pH(R−L+S) +C−I.
=pH(R−L+S) +C−I. (1)
Because of the assumption that the borrower can’t pledge future earnings due to political con- nections to the lender, I exclude S from Eq. (1) so that the expected total NPV including only
pledgeable income is
E[N P V|Hard, P ledgeable]T otal =pH(R−L) +C−I >0. (2)
I assume that if the entrepreneur works hard the E[N P V|Hard, P legeable]T otal > 0. If the en- trepreneur shirks, the expected total project NPV is
E[N P V|Shirk]T otal =pL(R−L+S) +B+C−I. (3)
I further assume that when only pledgeable income is counted
E[N P V|Shirk, P ledgeable]T otal =pL(R−L)−I <0. (4)
The restriction in Eq. (4) insures that the lender must design a contract that incentivizes the borrower to exert high effort and the assumption that Eq. (2) is positive insures that such a contract automatically satisfies the individual rationality constraint of the lender.
3 The influence of political connections on financing
3.1 The borrower and lenders contract
Using the payoffs from Table 1, I find the expected NPV to the lender when the borrower works hard is
E[N P V|Hard]Lender =epH(R−Rb) +C−(I−A). (5) Using the assumption that the expected profit to the lender is zero, I set Eq. (5) to zero. To find the borrowers contract, I then rearrange terms to find
Rb=R−(I −A)−C
epH . (6)
Eq. (6) shows the relationship between political connections and the borrower’s contract. First,
Rb is indepedent of both L and S. Intuitively, conditional on sufficent return R the lender is indifferent to the size of eitherLorS. To understand the influence of the other political connection parameters on the borrowers contract, I take the derivative of Eq. (6) with respect to eand C to find
∂Rb
∂e = 1
e2pH [(I −A)−C]>0, (7)
∂Rb
∂C = 1 epH
>0. (8)
Eq. (7) shows that as the enforcement probabilityeincreases, the payment to the borrower in the case of success Rb increases. Eq. (8) shows that if the political influence over the compensation to the lender C increases, then the payment to the borrower in the case of success with enforcement Rb also increases.
To find the lenders contract, I substitute Rl =R−Rb into Eq. (5), set the equation to zero, and rearrange to find
Rl= (I−A)−C epH
. (9)
Eq. (9) shows the relationship between political connections and the lender’s contract. As with the borrower’s contract, the lender’s contract is independent of bothL and S. To understand the influence of political connections on the lenders contract, I take the derivative of Eq. (9) with respect toeand C to find
∂Rl
∂e = C−(I−A)
e2pH <0, (10)
∂Rl
∂C = −1 epH
<0. (11)
As expected, the signs on the derivatives are opposite to the signs for the borrower’s contract.
Specifically, Eq. (10) shows that as enforcement eincreases(decreases) the payment to the lender in the case of success with enforcement decreases(increases). In response to a decrease in contract enforcement (i.e. e↓), the lenderex-ante increasesRl due to an increased probability of expropria- tion (1−e). Lastly, Eq. (11) shows C is a substitute for project return so an increase inC implies
a decrease inRl.
3.2 Interest rates and expected rate of return
Do political connections influence the interest rate of a loan? The contractual payment to the lender in the case of success with enforcement is Rl and the loan amount is I−A, which implies an interest rate of i= (I−A)Rl −1. In Appendix A.1, I solve forito find
i= 1 epH
1− C I−A
−1. (12)
Because I assume both risk neutrality and the risk free rate of interest equals zero,irepresents the rate of interest that exactly compensates the lender for the probability a state evenutuates where the lender is not paid.
Eq. (12) shows the interest rate is unaffected by eitherL orS. To understand the influence of political connections on interest rates, I take the derivative of Eq. (12) with respect to eand C to find
∂i
∂e = −1 e2pH
1− C I−A
<0, (13)
∂i
∂C = −1 epH
1 I−A
<0. (14)
Eq. (13) shows that as the enforcement probability e increases the rate of interest i decreases.4 If a politically connected borrower negatively influences the enforcement probability e, then Eq.
(13) shows the lender charges a politically connected borrower a higher interest rate. In contrast, Eq. (14) shows that an increase in political influence, which is manifested through an increase in the lender’s compensation C, results in a decrease in the rate of interest i. Thus, an increase in political influence both positively and negatively influences the rate of interest on the loan.
To see this more clearly, suppose a borrower that is not politically connected is subject to full enforcement e= 1 and can’t influence the compensation of the lender C = 0. Thus, the condition where a politically connected borrower has a higher rate of interest ip than the non-politically
4Note this relationship holds when we restrictC <(I−A).
connected borrowerin is
ip > in 1
epH
1− C I−A
−1> 1 pH −1, which I simplify to
e <1− C
I−A. (15)
Eq. (15) shows that a politically connected borrower (relative to a non-politically connected bor- rower) pays a higher interest rate when the enforcement probabilityeis low relative to the political influence of compensation scaled by the loan I−AC .
Do political connections influence the expected rate of return of the loan? To distinguish between the gross return R, I define the expected rate of return as E[r]. To match the model to the empirical literature, I follow Khwaja and Mian (2005) and define the expected rate of return as
E[r] = (1−d)(1 +i) +dc, (16)
wheredis the default rate andc the recovery rate. The model default rate dis
d= (1−e)pH + (1−pH)
d= 1−epH (17)
The model recovery rate isc= 0. I substitutec= 0, Eq. (17) and Eq. (12) into Eq. (16) to find
E[r] = (1−1 +epH)
1 + 1 epH
1− C I−A
−1
= 1− C
I−A. (18)
In Appendix A.2, I derive Eq. (18) in an alternative manner by taking the expected gross return to the lender divided the loan.5
5Note that Eq (18) defines the expected return of the loan and not the expected return of the lender, which is
Eq. (18) provides insight into the relationship between political connections and the expected rate of return of the loan. First, the expected rate of return E[r] is independent of both e,L, and S. This is becauseex-ante the lender writes a contract that accounts for possible expropriation. In addition, due to the zero NPV assumption the division of the surplus is not relevant to the lenders contract. Second, whenC= 0 the expected rate of return is equal to one, which is consistent with risk neutrality. Also, when the lender’s compensationC is influenced by political connections, the expected rate of return on the loan is less than one and furtherE[r] drops linearly based upon the ratio (I−A)C . In other words, the higher the compensation C relative to the loan amount (I−A) the lower the expected rate of return. Lastly, when C > 0 the negative expected rate of return equivalently implies a loan recovery rate that is less than the loan amount.
3.3 Access to credit
The conditions for the borrower to have access credit are: i) the project has sufficient pledgeable gross return to generate an expected positive NPV, ii) there must be sufficient return in the success with enforcement state to fund Rl, and iii) the entrepreneur must possess sufficient cash A to be incentive compatible. To clearly define the first condition, I re-arrange Eq. (2)
R > I−C pH
+L= ¯R1, (19)
where ¯R1 represents the minimum project gross return to receive financing.
The meet the second condition the borrower’s project must have sufficient gross return in the case of success with enforcement to meet the claim Rl. For example, in the extreme case of zero enforcement there would not be return available for the lender to receive Rl. Based on the zero profit condition to the lender,
epHRl= (I−A)−C. (20)
always equal to one. To see this, note that the lender’s return from the politically influenced lender compensation is
C I−A.
Also, for there to be sufficient total project return to payRl
R−L≥Rl
epH(R−L)≥epHRl (21)
I subsitute the RHS of Eq. (20) into the RHS of Eq. (21) and simplify to find
epH(R−L)≥(I−A)−C, R−L≥ (I−A)−C
epH , R≥ (I−A)−C
epH +L= ¯R2, (22)
where ¯R2 represents the minimum gross return required for the borrower to obtain funding.
The last condition is that the borrower has sufficient cash to be incentive compatible. The lender designs a contract so that
E[N P V|Hard]Ent≥E[N P V|Shirk]Ent
pH[e(Rb−L+S) + (1−e)(R−L+S)]−A≥pL[e(Rb−L+S) + (1−e)(R−L+S)] +B−A pH[eRb+ (1−e)R−L+S]≥pL[eRb+ (1−e)R−L+S] +B
∆p[eRb+ (1−e)R−L+S]≥B (23)
In Appendix A.3, I show that Eq. (23) implies that the minimum amount of cash required to secure a loan is
A¯=I−pH
R− B
∆p −L+S
−C (24)
For a borrower to obtain funding, she must have sufficient return so that R >R¯1 and R >R¯2 and sufficient cash so thatA >A¯as defined in Eqs. (19), (22), and (24), respectively. To understand the influence of political connections on access to credit, I evaluate each expression with respect to the parameters e,L,S, andC.
What is the influence of the enforcement probability e on obtaining financing? First, the
enforcement probability does not influence the minimum cash required or the gross return required for a positive NPV.6 However, a lower enforcement probability implies the borrower’s project must have a higher return due to the potential expropriation in the success but not enforce state. To clarify the relationship, I take the derivative of ¯R2 with respect toeto find
∂R¯2
∂e = −1
e2pH [(I−A)−C]<0, (25) which shows that return required decreases (increases) with the enforcement (expropriation) prob- ability.
What is the influence of the social objectives on financing? Because the borrower can not pledge to lender the present value of future from meeting social objectives,the return required to obtain financing is not affected byS.7 However, compensation due to meeting social objectivesSincreases the incentive of the borrower to work hard. To show this relationship, I take the derivative of ¯A with respect toS to find
∂A¯
∂S =−pH <0, (26)
which shows that increasing S implies a decrease in the cash required to secure financing.
In contrast, higher spending L on social objectives tighten financial constraints in all three dimensions. To more clearly see this relationship, I take derivatives of Eqs. (19), (22), and (24) with respect toL to find
∂R¯1
∂L = 1>0, (27)
∂R¯2
∂L = 1>0, (28)
∂A¯
∂L =pH >0. (29)
The signs of the derivatives show that an increase inL tightens financial constraints both in terms of the minimum gross returns and minimum cash required to obtain financing.
6Note that ∂∂eR¯1 = 0 and ∂∂eA¯ = 0.
7Note that ∂∂SR¯1 = 0 and ∂∂SR¯2 = 0.
What is the effect of political influence on the compensationCof the lender on access to finance?
To clarify the relationship, I take derivatives of Eqs. (19), (22), and (24) with respect to C to find
∂R¯1
∂C = −1 pH
<0, (30)
∂R¯2
∂C = −1 epH
<0, (31)
∂A¯
∂C =−1<0. (32)
The signs of the derivatives show that an increase inC loosens financial constraints both in terms of the minimum gross returns and minimum cash required to obtain financing.
The model provides insight into the effect of political influence on access to finance. First, a decrease in enforcement probability increases the minimum gross return R required to obtain financing. Second, an increase in the loss L due to meeting social objectives both increases the minimum gross return R and the minimum cash A required to obtain financing. Thus, at the margin a decrease ineand an increase in Lboth tighten access to credit. In contrast, an increase in political influence over the compensation C of the lender decreases the minimum cash A and returnRrequired to obtain credit. Overall, the influence of political connections on access to credit is dependent on relative magnitudes ofe,S,L, and C.
4 Why choose to be politically unconnected?
In practice, we observe networks of politically connected and unconnected borrowers. To explain why there is heterogeneity in political connections, one might argue that heterogeneity is due to the impossibility of everyone being politically connected, but social network theory suggests that if belonging to network is optimal, then everyone eventually joins the network.8 To evaluate if hetergeneity is an optimal strategy, I evaluate the conditions when a borrower chooses to become politically connected or unconnected. I conduct the analysis from the vantage point of when it would be beneficial for a borrower to be politically unconnected.
8For an overview of how information diffusion may result in the adoption of a common behaviour see Easley and Kleinberg (2010, Chap. 19).
As a starting point, a politically unconnected entrepreneur has no influence on the compensa- tion of the lender; hence C = 0 for a politically unconnected borrower. Likewise, the politically unconnected borrower’s future compensation is not tied to social obligations and so both L = 0 and S = 0. Because the borrower earns all the economic rents the expected NPV to the bor- rower is equal to the total project NPV.9 Thus, the NPVs to the politically connected borrower is E[N P V|Hard]p = pH(R −L +S) + C −I and to the politically unconnected borrower is E[N P V|Hard]n=pHR−I. A borrower would choose to be politically unconnected when
E[N P V|Hard]n≥E[N P V|Hard]p
pHR−I ≥pH(R−L+S) +C−I pH ≥ C
L−S (33)
Eq. (33) shows that a borrower chooses to be politically unconnected when the probability of success is high relative to the ratio L−SC . In the extreme case when political influence over lender compensation is so high thatC >(L−S) the borrower always chooses to be politically connected.10 Likewise, for C < (L−S) a relatively high C implies a borrower may choose to be politically connected even for a highpH. In contrast, for a very lowCa borrower may choose to be politically unconnected even for a relatively low pH. Likewise, as L−S increases a borrower may choose to be politically unconnected even for a relatively lowpH. Importantly, except in the case of extreme C the model suggests that a network of politically connected and unconnected borrowers co-exist.
Might a borrower choose to be politically unconnected to gain greater access to finance? More specifically, under what conditions does being politically unconnected imply the borrower needs either less gross return or cash to obtain financing. Thus, the conditions where a politically uncon- nected borrower has lower financial constraints are ¯R1n<R¯p1, ¯R2n<R¯p2, and ¯An<A¯p. In Appendix
9See Appendix A.4 for a derivation.
10Note that 0< pH<1<L−SC forC >(L−S).
A.5, I substitute the relevant expression into each condition to find
pHL > C, (34)
epHL+ (1−e)(I−A)> C, (35)
pH[L−S]> C. (36)
When the conditions in Eqs. (34) and (36) hold a politically unconnected borrower has less financing constraints than a politically connected borrower.11
Eq. (34) shows that a politically unconnected borrower (relative to a politically connected borrower) needs a lower gross return to obtain financing when the expected loss from social objec- tivespHL is greater than the compensation to the lenderC. Eq. (35) shows a second condition a politically unconnected borrower (relative to a politically connected borrower) needs a lower gross return to obtain financing. The RHS increases for two reason. First,epHLrepresents the expected loss in gross return in the success with enforcement state. Second, (1−e)(I −A) represents the expected value of the expropriation of the loan. When the sum of these effects are greater than C a politically unconnected borrower (relative to a politically connected borrower) requires less gross return Rto obtain financing. Lastly, Eq. (36) shows that a politically unconnected borrower (relative to a politically connected borrower) needs less cash to obtain financing when the expected loss from social objectivespHLminus the expected gain from social objectives pHS is greater than the compensation to the lender C. Note that if Eq. (36) holds, then Eq. (34) automatically holds.
Thus, the model shows that an entrepreneur with a relatively high probability of success may choose not to be politically connected either to fully capture the return of the project or secure financing. Because the model assumes symmetric information, the borrower is able to secure their preferred contract without a costly signal. The symmetric assumption is based on the idea the lender can observe if the borrower is (or is not) politically connected.12
11Note that Eq. (34) represents the condition that ¯Rn1 <R¯p1, Eq. (35) represents the condition that ¯Rn2 <R¯p2, and Eq. (36) represents the condition that ¯An<A¯p.
12If a borrower decides on a project by project basis to exercise or not exercise their political connection, then modeling the contract of a high expected return good type borrower from a lower expected return bad type borrower becomes an adverse selection problem similar to [pages 241-244]Tirole (2006).
5 Conclusion
In summary, I extend the fixed investment model with stochastic contract enforcement of Tirole (2006, pages 535-540), by adding politically influenced lender compensation, decrease project re- turn due to social costs, and implicit borrower compensation from meeting social objectives. In addition, I interpret the probability of contract enforcement as being negatively related to political connections. These additional features provide several insights. First, consistent with the empirical findings of Khwaja and Mian (2005) the model shows the expected rate of return on a loan is less than 100% when lenders receive politically influenced compensation. Second, loan interest rates increase as the probability of contract enforcement decreases, but decrease as politically influenced lender compensation increases. Third, the cost of social objectives counter-acts the influence over lender compensation and increases the required minimum project return and cash holdings required to obtain financing. Lastly, the model shows that a borrower with relatively high probability of project success is more likely to choose to be politically unconnected. In total, the model explains a number of empirical regularities.
The model leaves unanswered the related question of whether lenders should naturally separate into politically connected and unconnected banks. The empirical evidence regarding this question is mixed. The evidence in Khwaja and Mian (2005) shows that state banks in Pakistan tend to serve politically connected borrowers and private banks tend to serve unconnected borrowers. In contrast, the survey evidence of Allen, Qian, and Qian (2005) shows that state banks in China are an important early source of financing for private Chinese firms. However, the same survey evidence shows that state banks are uninvolved in financing firms during their expansion years. Rather, during these expansion years private firms used many sources of financing including friends, private credit agencies, trade credit, foreign direct investment, etc. Explaining the reasons underlying the reasons for these politically unconnected sources of finance, which exists outside of state banks, represents an interesting open question.
A Appendix
A.1 Derivation of Eq. (12)
I substitute Rl as defined in Eq. (9) into the expression below and simplify to find i= Rl
(I−A) −1
= 1
I−A
Rl−1
= 1
I−A
(I−A)−C epH
−1,
= 1
epH
1− C I−A
−1.
A.2 Derivation of Eq. (18)
Khwaja and Mian (2005) measure return as the percentage of the loan that is collected by the bank. In the model, this is equivalent to the expected gross return to the lender divided the loan.
I substitute Rl as defined in Eq. (9) into the gross return divided by the loan and simplify to find E[r] = epHRl+ (1−e)pH0 +pL0
(I−A) ,
= epHRl (I−A),
= epH
(I−A)−C
epH
(I−A) ,
= 1− C (I−A), which matches the derivation Section 3.2.
A.3 Derivation of Eq. (24)
I re-arrange Eq. (23) to
∆p[eRb+ (1−e)R−L+S]≥B eRb+ (1−e)R−L+S≥ B
∆p. (37)
I then substitute the contract to the borrower Rb =R−(I−A)−Cep
H into the first term in the LHS of Eq. (37) to find
e
R−(I −A)−C epH
+ (1−e)R−L+S ≥ B
∆p (38)
I then simply the Eq. (38) as follows:
e
R− (I−A)−C epH
+ (1−e)R−L+S≥ B
∆p eR−(I−A)−C
pH
+R−eR−L+S≥ B
∆p
−(I−A)−C pH
+R−L+S≥ B
∆p
−(I−A)−C pH ≥ −
R− B
∆p −L+S
−I+A+C≥ −pH
R− B
∆p −L+S
A≥I−pH
R− B
∆p −L+S
−C= ¯A A.4 Check that borrower receives all economic rents
The expected NPV to the entrepreneur is
E[N P V|Hard]Ent =epH(Rb−L+S) + (1−e)pH(R−L+S)−A (39) I substitute Rb =R−(I−A)−Cep
H into Eq. (39) and simplify to show E[N P V|Hard]Ent=epH
R− (I−A)−C epH
−L+S
−epH(R−L+S) +pH(R−L+S)−A,
=epH(R−L+S)−(I−A) +C−epH(R−L+S) +pH(R−L+S)−A,
=pH(R−L+S) +C−I, which equals the total project NPV.
A.5 Simplification of conditions in Eqs. (34), (35), and (36)
I simplify the condition in Eq. (34) to
R¯n1 <R¯p1 I
pH < I −C pH +L I < I −C+pHL
−pHL <−C pHL > C
I simplify the condition in Eq. (35) to
R¯n2 <R¯p2 (I−A)
pH
< (I−A)−C epH
+L e(I −A)<(I−A)−C+epHL
−epHL+e(I−A)−(I −A)<−C epHL−e(I−A) + (I −A)> C
epHL+ (1−e)(I −A)> C I simplify the condition in Eq. (35) to
A¯n<A¯p I−pH
R− B
∆p
< I−pH
R− B
∆p −L+S
−C
−pH
R− B
∆p
<−pH
R− B
∆p
−pH[−L+S]−C 0<−pH[−L+S]−C
pH[−L+S]<−C pH(L−S)> C
References
Allen, F., Qian, J., Qian, M., 7 2005. Law, finance, and economic growth in China. Journal of Financial Economics 77 (1), 57–116.
Berkman, H., Cole, R. A., Fu, L. J., 2009. Expropriation through loan guarantees to related parties:
Evidence from china. Journal of Banking & Finance 33 (1), 141–156.
Chen, D., Jiang, D., Ljungqvist, A., Lu, H., Zhou, M., 2015. State capitalism vs. private enterprise.
NBER Working Paper, 1–55.
Chen, S., Sun, Z., Tang, S., Wu, D., 2011. Government intervention and investment efficiency:
Evidence from china. Journal of Corporate Finance 17 (2), 259–271.
Easley, D., Kleinberg, J., 2010. Networks, crowds, and markets:Reasoning about a highly connected world, 1st Edition. Cambridge University Press, New York.
Khwaja, A. I., Mian, A., 2005. Do lenders favor politically connected firms? Rent provision in an emerging financial market. The Quarterly Journal of Economics 120 (4), 1371–1411.
Krueger, A. O., 1974. The political economy of the rent-seeking society. The American Economic Review 64 (3), 291–303.
Shleifer, A., Vishny, R. W., 1993. Corruption. The Quarterly Journal of Economics 108 (3), 599–
617.
Shleifer, A., Vishny, R. W., 1994. Politicians and firms. The Quarterly Journal of Economics 109 (4), 995–1025.
Tirole, J., 2006. The Theory of Corporate Finance, 1st Edition. Princeton University Press, Prince- ton, New Jersey.
Wei, L., Zhu, G., Sept. 13 2015. China unveils overhaul of bloated state sector: Beijing to allow state-owned enterprises to add private investors. Wall Street Journal Online.
